Ependymal cilia beating induces an actin network to protect centrioles against shear stress

HAL Id: hal-01822885

https://hal.sorbonne-universite.fr/hal-01822885

Submitted on 25 Jun 2018

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License

protect centrioles against shear stress

Alexia Mahuzier, Asm Shihavuddin, Clémence Fournier, Pauline Lansade,

Marion Faucourt, Nikita Menezes, Alice Meunier, Meriem Garfa-Traoré,

Marie-France Carlier, Raphael Voituriez, et al.

To cite this version:

Alexia Mahuzier, Asm Shihavuddin, Clémence Fournier, Pauline Lansade, Marion Faucourt, et al..

Ependymal cilia beating induces an actin network to protect centrioles against shear stress. Nature

Communications, Nature Publishing Group, 2018, 9, pp.2279. �10.1038/s41467-018-04676-w�.

�hal-01822885�

Ependymal cilia beating induces an actin network

to protect centrioles against shear stress

Alexia Mahuzier

1

, Asm Shihavuddin

1,6

, Clémence Fournier

2,7

, Pauline Lansade

1

, Marion Faucourt

1

,

Nikita Menezes

1

, Alice Meunier

1

, Meriem Garfa-Traoré

3

, Marie-France Carlier

2

, Raphael Voituriez

4,5

,

Auguste Genovesio

1

, Nathalie Spassky

1

& Nathalie Delgehyr

1

Multiciliated ependymal cells line all brain cavities. The beating of their motile cilia

con-tributes to the

flow of cerebrospinal fluid, which is required for brain homoeostasis and

functions. Motile cilia, nucleated from centrioles, persist once formed and withstand the

forces produced by the external

fluid flow and by their own cilia beating. Here, we show that a

dense actin network around the centrioles is induced by cilia beating, as shown by the

disorganisation of the actin network upon impairment of cilia motility. Moreover, disruption

of the actin network, or speci

fically of the apical actin network, causes motile cilia and their

centrioles to detach from the apical surface of ependymal cell. In conclusion, cilia beating

controls the apical actin network around centrioles; the mechanical resistance of this actin

network contributes, in turn, to centriole stability.

DOI: 10.1038/s41467-018-04676-w

OPEN

1_{Cilia biology and neurogenesis, Institut de biologie de l’Ecole normale supérieure (IBENS), Ecole normale supérieure, CNRS, INSERM, PSL Université Paris,}

75005 Paris, France.2Biochimie Biophysique et Biologie Structurale, I2BC, CNRS, 91198 Gif-sur-Yvette, France.3Cell Imaging Platform, INSERM US24 Structure Fédérative de Recherche Necker, Paris Descartes Sorbonne Paris Cité University, Paris, France.4Laboratoire de Physique Théorique de la Matière Condensée, CNRS/UPMC, Paris, France.5_{Laboratoire Jean Perrin, CNRS/UPMC, Paris, France.}6_{Present address: Department of Applied Mathematics and}

Computer Science, Technical University of Denmark, Kgs, Lyngby, Denmark.7_{Present address: UPMC University Paris 06 UM 1127, Inserm U1127, CNRS}

UMR7225, ICM, Sorbonne Université, Paris, France. These authors contributed equally: Nathalie Spassky, Nathalie Delgehyr. Correspondence and requests for materials should be addressed to N.S. (email:spassky@biologie.ens.fr) or to N.D. (email:delgehyr@biologie.ens.fr)

123456789

M

ulticiliated cells ensure the displacement of liquid or

mucus, which perform essential functions in the

organism, such as the displacement of oocytes in

fal-lopian tubes, the clearance of mucus from the airways, the stirring

of luminal

fluid in the efferent ducts or the circulation of

cere-brospinal

fluid (CSF) in brain ventricles

1,2

. Multiciliated cells in

the brain, called ependymal cells, line all brain ventricles and

form a protective barrier

3

. They also contribute to the neural stem

cell niche

4

. Ependymal cilia beating ensures the CSF

flow

necessary for brain homoeostasis, toxin washout, delivery of

signalling molecules and orientation of the migration of

new-born neurons

5

. Defective cilia motility is associated with

hydro-cephalus, which increases pressure in the skull due to an increase

in CSF in the ventricular cavities

6

. Ependymal cells are generated

from radial glial cells during early postnatal stages

7

. Their

regeneration during aging or under pathological conditions is

limited

8

, resulting in the partial loss of protection of the brain

parenchyma in aged patients

9,10

. Most ependymal cells persist,

however, throughout life.

Cilia beating results in mechanical constraints on the cells. For

example, in the multiciliated organism Tetrahymena thermophila,

if the ultrastructure of centrioles is abnormal, the forces exerted at

the base of beating cilia lead to centriole disassembly

11,12

. To

determine how centrioles withstand these mechanical constraints

lifelong, we focussed on the potential role of the actin

cytoske-leton

13

. The actin cytoskeleton of multiciliated cells has been the

focus of detailed analysis. Notably, in the multiciliated epidermal

cells of Xenopus larvae, it was demonstrated that two

inter-connected apical and subapical actin networks form a

3D-network that connects the centrioles, thus contributing to their

spacing and to the synchronisation of cilia beating

14

. The actin

cytoskeleton is assembled and organised progressively at the

apical surface of multiciliated progenitor cells

15–19

. It contributes

to the intercalation of the multiciliated cells in the Xenopus

epi-dermis

17,18

, guides the apical migration and docking of newly

formed centrioles in the plasma membrane

14–16,20,21

and

parti-cipates to ciliogenesis

16

. However, how this actin cytoskeleton

network is assembled in ependymal cells and what its direct links

with centriole stability are have not been addressed so far. In this

study, we describe the coordination between ciliary beating and

actin organisation that actively contributes to the protection of

ependymal cilia and centrioles against the shear stress generated

by ciliary beating and the associated

fluid flow.

Results

Actin assembles around centrioles during development. To

investigate when and how the actin network develops around

ependymal cell centrioles, lateral ventricular walls were

immu-nostained with antibodies against cilia, centrioles and

filamentous

actin (F-actin) at different stages during motile cilia formation.

The same area was analysed throughout the paper (Fig.

1

a).

F-actin is localised at cell borders at all stages (Fig.

1

a). In cells in

which centrioles are not yet ciliated (post-natal day 4, P4), actin

staining at the apical surface (defined by the localisation of the

distal centriolar marker FOP

22

) is scarce and diffuse. The staining

intensifies all over the cell cortex as ciliation begins (P6). As

motile cilia elongate during cell maturation, actin accumulates in

the centriolar patch (Fig.

1

a, b), and is positioned asymmetrically

at the front of mature ependymal cells

23

(P15). At this stage, the

actin network is thicker, extending from the apical surface of the

cells (apical actin) to ~1

μm below (subapical actin). Note that at

P15, the apical actin network consists of thick actin cables

oriented towards the rear of the cell, whereas the subapical

net-work contains smaller dot-like actin-positive structures in the

centriolar patch (P15, Fig.

1

a).

To characterise the morphology of the apical and subapical

actin networks connected to centrioles in mature ependymal cells

(P15), we used high resolution microscopy (STED, Fig.

1

c, d) and

3D-modelling (Fig.

1

e). Co-staining of actin with markers of the

distal appendages (Cep164

24

) or the distal lumen of the centrioles

(centrin

25

) revealed that the apical actin network is distributed

around the centrioles and extends from the distal appendages to

the centrin-positive distal portion of the centrioles (Fig.

1

c, f). We

occasionally observed apical actin

filaments surrounding clusters

of centrioles, the distal appendages of which were merged

(Fig.

1

d). Ependymal centrioles or clusters of centrioles are

anchored in the ciliary pocket, a depression of the plasma

membrane

26

, suggesting that the apical actin

filaments are

organised at this level. Co-staining of actin with markers of the

proximal end of the centrioles (Akap450

27

) and rootlets

(Rootletin

28

) shows that the subapical actin network extends

from the proximal end of the centrioles to their rootlets (Fig.

1

c,

f). The apical and subapical actin networks overlap partially along

the Z-axis and link together individual or clusters of centrioles

(Fig.

1

e). Therefore, in multiciliated ependymal cells, centrioles

are embedded in a three-dimensional actin structure from their

distal appendages to their rootlets. The assembly and organisation

of this complex actin network in the centriolar patch correlates

with the elongation of motile cilia (Fig.

1

a, b). It is unclear,

however, whether ciliogenesis induces the assembly of the actin

network at the centriolar patch or, conversely, whether the

increase in actin assembly leads to cilia elongation.

Cilia beating induces apical actin assembly at centrioles. To

assess the role of cilia in actin organisation, we studied mice in

which genes essential for the development of cilia, such as Kif3A

29

(a subunit of the motor protein kinesin-2) or Ift88

30

(a

homo-logue intraflagellar transport protein 88) or for cilia motility, such

as hydin

31

(that causes defects in the central pair of motile cilia

when mutated) or Lrrc6 (axonemal dynein subunit

32–34

) were

mutated or depleted.

We used hGFAP-Cre; Kif3A conditional knocked out mutant

mice (Kif3A-cKO) that have normal primary cilia in ependymal

cell progenitors, but are virtually devoid of motile cilia on the

lateral wall of the lateral ventricles

35

(Supplementary Figure

1

a).

F-actin

fluorescence immunostaining showed that actin was

distributed homogeneously on the apical cell surface in mature

Kif3A-cKO ependymal cells, whereas an additional apical actin

network was observed in the centriolar patch in controls (Fig.

2

a,

b). In the Ift88 conditional mutant (Ift88-cKO), similar results

were obtained (Supplementary Figure

1

b). Apical actin

enrich-ment at the centriolar patch correlates with cilia elongation

(Fig.

1

a, b) and two independent mutant mice in which motile

cilia are absent, lack the apical centriolar actin network (Fig.

2

a, b;

Supplementary Figure

1

b). Thus, these results indicate that motile

cilia are a pre-requisite for actin enrichment at the centriolar

patch.

Unlike the apical actin network, the subapical centriolar

distribution of actin was identical in the wild-types and the

ciliary mutants (Fig.

2

a, c; Supplementary Figure

1

b). Therefore,

the organisation of the subapical actin network around centrioles

is independent of both motile cilia and enrichment of the apical

actin network in the centriolar patch. Since both ciliary mutant

mice (Kif3A-cKO and Ift88-cKO) are devoid of motile cilia, it is

unclear whether the cilia themselves or their motility is required

for assembly of apical actin around the centrioles. To determine

whether cilia motility is required, we used hydin mutant mice that

have stiff vibrating cilia

31

. Immunostaining of the cilia in hydin

mutant mice confirmed that mature (P15) ependymal cells extend

cilia that are normal in terms of length and density

Apical actin enrichment at centrioles (centriolar patch/ remaining surface)

Apical

c

d

Actin

Cep164

Actin Cep164 Akap450

Actin Rootletin

b

Actin Akap450 Apical Subapical

Apical actin Subapical actin

Actin Centrin P15 Akap450 Cep164 Centriole Actin Proximal markers 3D modelling STED STED P15 Distal markers A P V D Centrin Rootletin

e

Actin Cep164 Centrin Actin Centrin Apical Subapical Cilium

f

z x y

***

***

Apical actin Subapical actin

P15 Cilium

a

P6 P15 P4 Centrioles Cilia Cilia <5 µ m Mature, cilia >5 µ m N ot ciliated Centrioles Apical Subapical 0.0 0.5 1.0 1.5 2.0 2.5 Actin Actin No cilia <5 µm >5 µm Cilia length

Fig. 1 Formation of a dense actin network at the centriolar patch during ependymal cell differentiation a F-actin (phalloidin, grey), centrioles (FOP, red) and cilia (GT335, green) in whole mounts of lateral ventricular walls at postnatal days 4 (P4), P6 and P15 at the apical level (colocalising with FOP staining) and subapical level; the actin network in the centriolar patch thickens as motile cilia grow; cell borders and centriolar patches are outlined with dashed white and red lines, respectively. Arrows point to long actinfilaments. On the right, a sagittal view of the lateral ventricular wall and the region of interest analysed throughout the paper (red square); A: Anterior, P: Posterior, D: Dorsal, V: Ventral.b Ratio of mean phalloidinfluorescence intensities in the centriolar patch (red dashed lines ina) and on the remaining cell surface (white minus red dashed lines) as a function of ciliary length. Error bars represent the sem of n = 112 cells from three mice; P-values were determined by one-way ANOVA followed by Dunn’s multiple comparison test; ***P ≤ 0.0001. c–e High resolution microscopy (STED) of mature ependymal cells on P15 whole-mount lateral ventricular walls.c Staining of F-actin (phalloidin, grey) and centriole distal markers, Cep164 (distal appendages, red) or Centrin2-GFP (distal lumen, yellow) that colocalise with the apical actin network, or centriole proximal markers Akap450 (pericentriolar material, green) or Rootletin (rootlets, cyan) that colocalise with the subapical actin network.d Zoom-in of Cep164 (red)-, CentrZoom-in2-GFP (yellow)- and F-actZoom-in (phalloidZoom-in, grey)-labelled mature ependymal cells; note a cluster of two centrioles with merged distal appendages (arrowhead) surrounded by actin.e Three-dimensional modelling of Cep164 (red)-, Akap450 (green)- and F-actin-(grey)-positive centrioles showing the links between the apical and subapical networks, and between neighbouring centrioles.f Model of centrioles with their associated actin bundles positioned according to the markers used in this study. Scale bars=5 μm in a and 1 μm in c, d

(Supplementary Figure

1

a, c). Line scans (kymograms) and

quantification of cilia beating frequency on side views of videos of

motile cilia showed that the beating frequency was reduced in

hydin mutant mice (Fig.

2

d) and its amplitude abnormal

(Supplementary Figure

1

d), confirming that hydin mutant cilia

are stiff and beat slowly

31

. F-actin immunostaining in mature

ependymal cells in hydin mutants showed that the subapical actin

network organisation was unchanged compared to control cells

(Fig.

2

a, c), but the apical actin enrichment in the centriolar patch

was decreased (Fig.

2

a, b), suggesting that cilia motility is required

for the apical enrichment of actin around the centrioles of mature

ependymal cells.

To confirm whether cilia motility is required for the

enrichment in apical actin at the centriolar patch, we blocked

cilia motility by using an shRNA against Lrrc6, a subunit of

axonemal dynein required for this motility

32–34

. We validated this

strategy by the observation of a decrease in the frequency of cilia

beating with high speed video microscopy on Lrrc6-depleted

ns ns ns ns Beating frequency (Hz) WT Kif3A cKO Centrioles Hydin

Apical actin Subapical actin

P15 NiCl 2 1 mM WT Hydin NiCl2

***

***

_***

_***

Depletions Subpical actin enrichment at centrioles

(centriolar patch/remaining surface)

PBS

Explants

shLrrc6

shCtrl

Depletions

Mutants Mutants Explants

Depletions Explants Mutants 50 ms Brain slices Cell culture shCtrl shLrrc6

Brain slices Cell culture

Apical actin

enrichment at centrioles

(centriolar patch/remaining surface) ₀ 1 2 3 WT Kif3A cKO Hydin shCtrl shLrrc6 PBS 0 WT Kif3A cKO Hydin shCtrl shLrrc6 PBS 1 2 3 WT 0 20 40 60 Hydin shCtrl shLrrc6 NiCl₂ 1 mM NiCl2 1 mM

***

***

***

NiCl2 1 mM

a

b

c

d

ependymal cells in culture

36,37

(Fig.

2

d). Electroporation of the

shRNA construct in the lateral ventricles of new-born mice

37,38

and immunostaining of cilia at P15 showed that Lrrc6 depletion

did not affect cilia length and density (Supplementary Figure

1

a,

c). Analysis of F-actin immunostaining at P15 revealed that, in

absence of Lrrc6, apical actin was not enriched at the centriolar

patch whereas the organisation of subapical actin remained

unaffected (Fig.

2

a–c). We also measured F-actin fluorescence

intensity at the apical and subapical levels compared to

surrounding none-transfected cells, and showed that both apical

and subapical actin levels are decreased in Lrrc6-depleted mature

ependymal cells (Supplementary Figure

1

e, f). Altogether, our

results suggest that cilia motility is required for the enrichment of

apical actin at the centriolar patch and for the assembly of a

subpopulation of actin cables at the apical and subapical levels.

To assess whether cilia beating is also required to actively

maintain the apical actin enrichment at the centriolar patch, we

blocked cilia beating in explants of the lateral ventricular wall of

P15 mice, which are lined with ependymal cells bearing motile

cilia and fully developed actin networks. To block cilia beating, we

used NiCl

2

, which has a direct effect on axonemal dynein

11,12,39

.

Video-recording of treated cilia confirmed that the cilia stopped

beating in the treated cells (Fig.

2

d; Supplementary Figure

1

d),

whereas cilia length was unaffected (Supplementary Figure

1

a, c).

F-actin immunostaining showed, however, a significant decrease

in apical but not subapical actin enrichment at the centriolar

patch (relative to control explants treated with PBS), when cilia

motility was stopped (Fig.

2

a–c).

Thus, using genetics or pharmacological treatments, we show

that cilia motility is required to initiate and maintain the apical

actin enrichment at the centriolar patch.

Actin maintains centrioles in mature ependymal cells. To assess

the role in centriolar function of actin assembled around

cen-trioles, we treated P15 explants of lateral ventricular wall, lined

with ependymal cells bearing motile cilia and assembled actin

networks, with cytochalasin-D, which progressively

depoly-merised the actin network at both the apical and subapical levels

and at cell borders, in a dose- and time-dependent manner

(Fig.

3

a). Neither the length of the motile cilia (Supplementary

Figure

2

a, b) nor their beating frequency (Fig.

3

b) was

sig-nificantly affected, even by a long treatment with a high

con-centration (2

μM) of cytochalasin-D. However, cells, cilia tufts

and centriolar patches were smaller after cytochalasin-D

treat-ment (Fig.

3

a; Supplementary Figure

2

a, c) and the

inter-centriolar distance was slightly decreased (Supplementary

Fig-ure

2

d), as previously described in multiciliated Xenopus cells

14

.

In addition, the number of centrioles also decreased significantly

in cytochalasin-D-treated explants (Fig.

3

c, d). This decrease was

time- and dose-dependent (Fig.

3

c, e, f). The decrease in the size

of the centriolar patch observed after actin disruption

(Supple-mentary Figure

2

c) was thus due to both the loss of centrioles and

the decrease in the inter-centriolar distance. Altogether, actin

contributes to centriole maintenance and spacing in ependymal

cells.

The apical actin contributes to centriole maintenance. To

determine which actin network, apical or subapical, is required

for centriole maintenance, we sought of actin nucleators

specifi-cally involved in these networks. There are three different families

of actin nucleators, the Arp2/3 complex that nucleates branched

actin

filaments and the formins and WH2 (Wiskott Aldrich

syndrome protein homology 2)-containing proteins that both

nucleate linear actin

filaments

40

. We treated explants of lateral

walls of P15 animals, lined with ependymal cells bearing motile

cilia and assembled actin networks, with specific inhibitors of the

Arp2/3 complex (CK666

41

) or formins (SMIFH2

42

) to assess their

specific contribution to the actin structures of mature ependymal

cells. Since we did not observe obvious effect on actin

organisa-tion after CK666 treatment (12 h at 100μM, Fig.

2

e), we focused

on formins. With commercially available antibodies, we found

several formins, namely Fhod1, Fmn1, Daam1 and mDia1 that

co-localised with centriole markers in ependymal cells (Fig.

4

a),

as previously shown in multiciliated Xenopus epidermal cells for

Daam1 and Fmn1

17,43

. Interestingly, compared to cytochalasin-D

treatment that affected both the apical and subapical actin

net-work (Fig.

3

a), SMIFH2-treatment led to a strong decrease of the

apical actin enrichment in the centriolar patch with no major

effect on the organisation of the subapical actin network

(Fig.

4

b–d). These results suggest that formins contribute at least

to the maintenance of the apical actin enrichment at the

cen-triolar patch. Interestingly, unlike the effect of cytochalasin-D

treatment, no difference in the inter-centriolar distance was

observed after SMIFH2-treatment (Supplementary Figure

2

f),

suggesting that the distance between centrioles is, as previously

suggested

14

, mainly regulated by the subapical actin organisation

(compare Supplementary Figs.

2

d and 2f). Formin inhibition as

well as actin depolymerisation with cytochalasin-D did not affect

cilia length or beating (Supplementary Figure

2

a, b; Fig.

4

e).

Noteworthy, SMIFH2-treatment induced smaller centriolar

pat-ches (Supplementary Figure

2

g) with decreased number of

cen-trioles relative to DMSO treated explants (Fig.

4

f, g). Altogether,

these results suggest that, in mature ependymal cells, apical actin

in the centriolar patch helps maintain the correct number of

centrioles.

To confirm that the apical actin network is implicated in

centriole maintenance, we examined the role of the actin binding

protein Cordon-Bleu (Cobl), another actin regulator involved in

the establishment of apical actin structures such as microvilli

44–

46

_{. Cobl nucleates and severs linear actin}

_{filaments in vitro and is}

Fig. 2 Cilia beating is required for the formation and maintenance of the apical actin network at the centriolar patch a F-actin (phalloidin, grey) and centrioles (FOP, red) in mature ependymal cells (P15) in ciliary mutant mice (Kif3A cKO or hydin) and wild-type (WT) mice transfected with shRNA (shCtrl, Control or shLrrc6 directed against the axonemal dynein subunit Lrrc6) or explants of WT ventricular walls treated for 12 h with PBS or 1 mM NiCl2

to block cilia beating; cell borders and centriolar patches are outlined with dashed white and red lines, respectively. Apical actin enriched in the centriolar patch is reduced in ciliary mutants and after NiCl2treatment, whereas the subapical network appears unaffected. Scale bar=5 μm. b, c Ratio of mean

phalloidinfluorescence intensities in the centriolar patches (red dashed lines) and in the rest of the cell surface (white minus red dashed lines) at the apical (b) and subapical (c) levels. Error bars represent the sem of: at the apical level, n = 150 cells for WT, 85 for Kif3A-cKO and 103 for hydin mutants; 25 for shCtrl, 32 for shLrrc6, 53 for PBS- and 122 for NiCl2-treated explants; at the subapical level, n = 91 cells for WT, 50 for Kif3A-cKO and 51 for hydin mutants;

22 for shCtrl, 32 for shLrrc6,50 for PBS- and 55 for NiCl2-treated explants; from three independent experiments; P-values were determined by one-way

ANOVA followed by Dunn’s multiple comparison test; ***P ≤ 0.0001; ns, not significant P > 0.05. d Representative kymographs (200 ms) of motile cilia in brain slices from control (WT) or hydin mutant mice or an explant from a WT ventricular wall treated for 12 h with 1 mM NiCl2or of motile cilia of

ependymal cells in culture transfected with shCtrl or shLrrc6. The graph shows the cilia beating frequency (Hz). Error bars represent the sem of n = 40 cells for WT and hydin mutants; n = 54 cells for NiCl2-treated explants, n = 62 cilia for shCtrl; n = 47 cilia for shLrrc6; from three independent experiments;

mostly expressed in mammalian brain tissue

47,48

and in ciliated

epithelial cells in zebrafish

49

_{. We}

_{first verified that Cobl is}

expressed in differentiating ependymal cells by semi-quantitative

RT-PCR in different phases of ependymal cell differentiation in

culture

36,37

. Increased expression of Cobl correlated with an

increase in the number of multiciliated cells (Supplementary

Figure

3

a). In the absence of an antibody against mouse Cobl, we

localised the Cobl protein in ependymal cells by electroporating

the lateral ventricular walls in new-born mice

37,38

with a

Cobl-DsRed construct and determined its location at P15. F-actin

labelling in mature ependymal cells showed that the Cobl-DsRed

is expressed in the apical but not subapical actin network

(Fig.

5

a).

We then examined, whether Cobl depletion disrupts the apical

actin network during ependymal cell development by

electro-porating an shRNA against mouse Cobl in the brain ventricles at

birth. Two shRNA were designed and validated (Supplementary

Figure

3

b–d, and see below). In immature ependymal cells (P6),

no defects in actin organisation or centriole formation, migration

or docking were observed in Cobl-depleted cells, showing that

Cobl is not needed in early stages of multiciliated ependymal cell

development (Supplementary Figure

3

e–i). In mature ependymal

cells, however, Cobl-depletion significantly decreased actin levels

relative to control shRNAs (Fig.

5

b–d), and more highly in the

apical compared to subapical level. Cobl is, therefore, required for

the maintenance of the apical actin network in mature ependymal

DMSO Centrin-GFP 40 h P15 2 µM Centrin-GFP Time Dose 50 ms DMSO CytoD 40 h Beating frequency (Hz) ns

Fold change in centriole number

***

Centriole number

*

*

***

***

***

Fold change in centriole number

DMSO 0 20 40 60 CytoD 2 µm DMSO 0 50 100 DMSO CytoD 2 µm 0.5 0.7 0.9 1.1 CytoD 1 µm 0.5 0.7 0.9 1.1 CytoD 12 h DMSO control 40 h

a

Centrioles _{Apical actin} P15

Actin depolymerizing drug CytoD 1 µM 40 h

Centrioles _{Apical actin}

Subapical actin

Subapical actin

Actin depolymerizing drug CytoD 2 µM 12 h

Centrioles _{Apical actin}

Subapical actin

Actin depolymerizing drug CytoD 2 µM 40 h

Centrioles _{Apical actin}

Subapical actin CytoD 1 µM 40 h CytoD 2 µM 40 h CytoD 2 µM 12 h

***

***

***

DMSO CytoD 6 h CytoD 40 h CytoD 40 h 2 µm

c

b

d

e

f

cells. Inter-centriolar distance was unaffected by Cobl depletion

(Supplementary Figure

4

a), consistent with its localisation at the

apical level, confirming that centriolar spacing mainly depends on

the subapical actin network. Ependymal cilia beating was still

efficient after Cobl depletion, even though the motile cilia were

slightly shorter (Supplementary Figure

4

b, c), in agreement with a

previous report

49

. Indeed, centriole orientation, which is known

to depend on cilia beating

37,50

, was not randomised in

Cobl-depleted cells in vivo (Supplementary Figure

4

d, e). Similarly, cilia

still beat at a high frequency in Cobl-depleted cells in vitro, only

slightly lower than in controls probably due to their shorter

length (Supplementary Figure

4

f). Remarkably, centriolar patch

size and centriole number were significantly decreased in

Cobl-depleted cells (Fig.

5

e, f; Supplementary Figure

4

g). All centrioles

were located in the apical plasma membrane stained with the

membrane-Cherry electroporation reporter, suggesting that no

centrioles were released into the cytoplasm (Fig.

5

g). To note, the

ciliary tuft size is affected to the same extent that the centriolar

patch size by Cobl-depletion (Supplementary Figure

4

b, g),

suggesting that cilia are lost with centrioles when actin is

impaired.

These results show that both Cobl and formins contribute to

the enrichment of apical actin at the centriolar patch and the

maintenance of centrioles.

The apical actin protects cilia/centrioles from detachment.

Although, apical actin may be required for centriole maintenance,

we noted that centriole loss did not occur in ciliary mutants

(either Kif3A-cKO or hydin) (Supplementary Figure

5

a, b) even if

actin enrichment is decreased (Fig.

2

a, b). We thus hypothesised

that the forces exerted by cilia beating may be responsible for the

loss of centrioles in the absence of apical actin. We used two

different strategies to rescue centriole loss in the absence of apical

actin. We

first depolymerised actin in P15 explants with

cyto-chalasin-D, alone or in association with NiCl

2

that blocks cilia

beating. Actin was affected by cytochalasin-D treatment in the

presence as in absence of NiCl

2

(Supplementary Figure

5

c).

However, centriole loss was prevented if both drugs were

com-bined (Fig.

6

a, b). For the second strategy, we depleted Cobl in

Kif3A-cKO cells, devoid of motile cilia. This depletion induced a

decrease of apical actin (Supplementary Figure

5

d, e) without

affecting centriole number (compare the number of centrioles

after Cobl depletion in control Fig.

5

e, f or in Kif3A-cKO Fig.

6

c,

d). Thus, the apical actin is required to protect centrioles only

when motile cilia are fully active, suggesting that the forces

exerted by the movement of cilia and/or the external

fluid flow

leads to centriole loss when the apical actin is disrupted.

Since no fragmented or whole centrioles were detected inside

the cells when the apical actin network was impaired, we

hypothesised that the centrioles might have been expelled from

cells. To test this hypothesis, we collected the supernatant of

mature ependymal cells in culture incubated with DMSO

(controls) or cytochalasin-D, alone or in association with NiCl

2

.

Centrioles with their associated cilia were frequently observed in

the supernatants of cytochalasin-D-treated cells compared to the

DMSO-treated controls or cells in which cilia motility was

inhibited by NiCl

2

(Fig.

6

e, f). This suggests that when the actin

network is destabilised, the forces generated by cilia beating cause

the centrioles to detach from the apical membrane. To assess

whether the

flow induced by cilia beating is sufficient to expelled

centriole/cilium from the ependymal cells, we applied an artificial

external

fluid flow

37

, comparable to physiological

flow, on cells

treated with cytochalasin-D and NiCl

2

. Since the number of

centriole/cilium found in the supernatant was not significantly

increased compared to controls, it is likely that the shear stress

induced by the movement of the cilium itself promotes centriole

destabilization in ependymal cells. Nevertheless, we cannot

exclude a weak contribution of the collective

flow on the

expulsion.

When apical actin is decreased (either with drugs or in

Cobl-depleted cells) cilia seem to be lost, as well as centrioles

(Supplementary Figs.

2

a,

4

b). No centriole or cilium are observed

within the cell. The presence of centriole/cilium in the

super-natant of ependymal cells treated with cytochalasin-D suggest

that centrioles are expulsed from the cell through the shear stress

generated by cilia beating forces, when actin is impaired.

Alternatively, drug-induced actin depolymerisation might lead

to a centriole/cilium detachment from the cell, while

physiolo-gically actin protects centrioles against destabilisation through

other mechanisms. Altogether, our results show that cilia motility

induces the assembly of a dense apical actin network around

centrioles that, in turn, contributes to centriole stabilisation

against forces developed by cilia beating/fluid flow (Fig.

6

g).

Discussion

We have demonstrated that, during the maturation of ependymal

cells in the developing brain, actin assembles around centrioles at

the base of cilia in two functionally different apical and subapical

networks. Enrichment of apical actin around the centrioles is

induced mechanically by cilia beating. Cilia beating produces

different forces that counteract (i) the force of CSF

flow (F

flow

)

and (ii) the additional torque induced by cilia movement (F

torque

).

To determine an order of magnitude for these forces, we

sim-plified the system by considering cilia as cylinders with the

geo-metrical parameters of motile cilia (measured on electron

micrographs or on CD24 immunofluorescence), such as radius, r

= 99 nm ± 2 (mean ± sem), centriole length, L

c

= 510 nm ± 2

(mean ± sem) and cilia length, L

= 11.1 μm ± 0.1 (mean ± sem)

Fig. 3 Actin is required for centriole stability in mature ependymal cells. a F-actin (phalloidin, grey) and centrioles (FOP, red) in mature ependymal cells from explants treated with DMSO or 1 or 2μM cytochalasin-D (CytoD) for 12 or 40 h showing concentration- and time-dependent actin impairment; cell borders and centriolar patches are outlined with dashed white and red lines, respectively.b Representative 200 ms kymographs of motile cilia in brain slices treated for 40 h with DMSO or 2_{μM cytochalasin-D. The graph displays the ciliary beating frequency (Hz) that was not affected by the drug. Error} bars represent the sem of n = 48 cells for DMSO- and CytoD-treated explants, from three independent experiments; P-values were determined with the Mann-Whitney test; ns, P > 0.05. c Representative images of Centrin2-GFP+centriolar patches in mature ependymal cells treated with DMSO or 1 or 2μM cytochalasin-D for 12 to 40 h.d Representative quanti_{fication of the number of centrioles per cell in lateral wall explant at P15 with contralateral brain of the} same animal treated with DMSO or cytochalasin-D at 1μM for 40 h. Error bars represent the sem of n = 301 cells for DMSO- and n = 93 for CytoD-treated explants, from one representative animal out of 6; P-values were determined with the Mann–Whitney test; ***P ≤ 0.0001. e–f Fold change in centriole number relative to DMSO-treated contralateral brain in lateral ventricles treated with 1 or 2_{μM cytochalasin-D for 6 up to 40 h. A time- (e) and dose- (f)} dependent reduction in the number of centrioles was observed following actin depolymerisation; P-values were determined by one-way ANOVA followed by Dunn’s multiple comparison test. Error bars represent the sem of for the dose dependent experiments, n = 389, 388 and 940 cells for DMSO-, CytoD 1µM- and CytoD 2µM-treated explants, respectively, from at least three independent experiments. For the time-dependent experiments, n = 2852, 704, 587 and 889 cells for DMSO-, CytoD 6h-, CytoD 12h- and CytoD 40 h-treated explants from at least three independent experiments. *** P ≤ 0.0001. * DMSO vs. CytoD 6 h P = 0.0287. *CytoD 12 h vs. CytoD 40 h, P = 0.0484. Scale bars=5 μm

(Supplementary Figure

6

). These cylinders are considered to be

immersed in a medium of low viscosity with their base embedded

in a visco-elastic medium simulating the actin network.

According to a previous calculation

51

, the force counteracting the

hydrodynamic

flow may be expressed as F

flow

= CfAL, where L is

the length of the cilia, f the beating frequency (Fig.

2

d), A the

beating amplitude (Fig.

2

e) and C is the normal force coefficient

(C=

4πμ

LogðL=rÞ

= 6×10

−3

, where

μ=10

−3

Newton s m

−2

is the

visc-osity of the CSF

52

, and r is the radius of the cilium). Thus, the

force developed by a beating cilium at its base to counteract the

fluid flow is on the order of F

flow

= 33pNewton (pN). Cilia

beating also induces an additional torque, i.e., a shear stress on

the centriole. In response to this torque a higher force, local and

temporary, is generated at the base of cilium that can be estimated

Apical actin

Apical actin

Formin inhibitor SMIFH2 60 µM

Subapical actin Subapical actin ns 50 ms DMSO SMIFH2 DMSO control Centrioles 12 h Centrioles P15 Beating frequency (Hz)

*

***

Daam1 Fmn1 Fhod1 P15 mDia1 Fold change in centriole number Centrin-GFP 40 h P15 SMIFH2 60 µM DMSO Merge Formins 0 1 2 3 0 1 2 3 DMSO 0.5 0.7 0.9 1.1 DMSO 0 20 40 60

Fhod1 Centrioles mDia1 Centrioles Fmn1 Centrioles

Daam1 Centrioles

Apical actin enrichment at centrioles (centriolar _{patch/remaining surface)}

Subapical actin enrichment at centrioles (centriolar patch/remaining surface)

SMIFH2 60 µm DMSO SMIFH2 60 µm DMSO SMIFH2 60 µm

***

SMIFH2 60 µm

a

b

c

d

e

f

g

from the equation F

torque

=F

flow

L/L

c

, in which L

c

is the size of the

centriole stabilised by the actin-rich network. Thus, F

torque

is on

the order of 728pN. Interestingly, this force is similar to those

developed by some

flagella for propulsion

53

. We have already

determined that some formins (mDia1, Fmn1, Daam1 and

Fhod1) and WH2-dependent actin nucleator (Cobl) are

impli-cated in the assembly of the apical network. The contribution of

formins is particularly noteworthy. Indeed, some members

of the formin family, such as mDia1, have been shown to be

mechanosensitive

54–57

. Moreover, formin inhibition (using

SMIFH2) led to a specific decrease of the apical actin enrichment

at the centriolar patch reminiscent of the actin phenotype

observed in ciliary mutant or after chemical inhibition of ciliary

motility. Further studies are needed to definitely demonstrate

that formins are indeed activated by ciliary beating in

multi-ciliated cells.

In turn, the apical actin network protects centrioles against the

shear stress, generated by the

fluid flow and/or by mechanical

torsion. Our in vitro data using drug to induce actin disruption

suggest that actin protects centriole by preventing them to be pull

out from the cell by motile cilia. However, we cannot rule out

other more direct role of actin for protecting centrioles. The

direct molecular actors linking centrioles to actin have not been

investigated in this study. In this respect, it would be of interest to

analyse the role of centriolar actors also involved in the

main-tenance of centrioles, such as Poc1, FOP or Cep135

11,12,58

. The

feedback loop between cilia beating and the structural

organisa-tion of the centriolar patch controlled by actin, preserves the

appropriate number of motile cilia in ependymal cells and thus

maintains efficient CSF flow in the brain ventricles. This

main-tenance is essential as the CSF contains many signalling

mole-cules that actively regulate mammalian brain development,

behaviour (e.g. sleep, appetite) and responses to injury

5

.

More-over, impairment of the CSF

flow has also been correlated with

age-related dementia

59

and hydrocephalus.

Subapical actin might also help maintain the centrioles, since it

was slightly decreased in most experiments. A close relationship

between centrioles and the subapical actin network has been

noted previously

43

, suggesting that the reduction of the

enrich-ment of the subapical actin network observed in our experienrich-ments

might be due mostly to the decreased number of centrioles. The

subapical actin network may be required for the correct spacing

of centrioles, as in multiciliated Xenopus epidermal cells

14

,

although the reduction in the inter-centriolar distance might also

be influenced by cell shrinkage. Indeed, the inter-centriolar

dis-tance was affected by cytochalasin-D that disrupts both the apical

and subapical actin networks, but not after SMIFH2-treatment or

Cobl depletion that mainly disrupt apical actin enrichment at the

centriolar patch.

A role of actin in the maintenance of centrioles has not been

observed thus far in other multiciliated tissues or organisms

14,60

.

This could be due to insufficient depolymerisation of the apical

actin network in previous studies, since we have shown that the

disappearance of centrioles depends on the dose and duration of

treatment with the actin-depolymerising agent.

Many questions remain to be answered such as: What is the

detailed structural organisation of the apical and subapical actin

networks? What nucleators initiate the subapical network? What

roles are played by each of the various formins localised at the

apical region of centrioles? Are these formins mechanically

acti-vated by cilia beating? What specific roles are played by the

formins and Cobl, or what signalling pathways activate them?

Interestingly, the role of RhoA-activated actin assembly by Fmn1

in the emergence of the apical surface of multiciliated cells has

recently been reported

17,18

.

Methods

Transgenic mice. All animal studies were performed in accordance with the guidelines of the European Community and French Ministry of Agriculture and were approved by the Ethic comity Charles Darwin (C2EA-05) and“Direction départementale de la protection des populations de Paris”, (Approval number Ce5/ 2012/107; APAFiS #9343). The mouse strains have already been described: RjOrl: SWISS (Janvier labs), Cen2–GFP61_{(CB6-Tg(CAG-EGFP/CETN2)3-4Jgg/J, The}

Jackson Laboratory); B6-Kif3atm2Gsn_{; B6-Ift88}tm1Bky30_{, C57BL/6-Tg(hGFAP-Cre)} 62_{, C57BL/6-Tg(Nestin-Cre)}63_{, hydin (B6CBACa Aw-J/A-Hydin hy3/J, The}

Jack-son Laboratory64). To produce hydin homozygous mutants, we crossed hydin+/− or hydin+/−/ Cen2-GFP with hydin+/−or hydin+/−/ Cen2-GFP. To produce the conditional mutant Kif3a (Kif3A-cKO), we crossed Kif3afl/flor Kif3afl/fl/ Cen2–GFP mice with Kif3ako/+/ Tg(hGFAP-Cre) mice. To produce the conditional mutant Ift88 (Ift88 cKO), we crossed Ift88fl/flmice with Ift88ko/+/ Tg(Nestin-Cre). Wild-type littermate mice were used as controls.

Plasmids. Cobl, derived from the mouse Cobl complementary DNA clone mKIAA0633 (Kazusa DNA Research Institute, Japan), was cloned in pDsRed-N1 (Clontech). To determine siRNA sequences to obtain Cobl depletion, we screened ten siRNA sequences against Cobl-GFP ectopically expressed in Hela cells. Two were selected (Supplementary Figure3b): Cobl sh1:

5′ACCGGTGCATGGTTC-TAGTCACATTCAAGAGATGTGACTAGAACCATGCACTTTTTCTCGAGG3′ and Cobl sh2: 5 ′ACCGGTCATCGACTCAAGACTATTTCAAGAGAA-TAGTCTTGAGTCGATGACTTTTTCTCGAGG3′. We also selected an shRNA corresponding to an Lrrc6 shRNA previously designed on human sequences from a homologous region32_{: 5}

′ACCGGTGCCCAAGGTAGGAGAAATGATTTCAA-GAGAATCATTTCTCCTACCTTGGGCACTTTTTCTCGAGG3′. We cloned the shRNA sequences in the shRNA backbone vectors pcDNA-U6-shRNA and pTRIP-U6-shRNA-IRES-PGK-GFP-WPRE, as previously described65_{. The control shRNA}

(shCtrl) was generated from AllStars Negative Control siRNA (Qiagen), which does not recognise mammalian RNAs. The reporter plasmid encoding membrane Cherry (mCherry) under the control of the CAGGS promoter was co-transfected with pcDNA-U6-shRNA and was a gift from Xavier Morin. All cells expressing the shRNA-GFP (pTRIP-U6-shRNA-IRES-PGK-GFP-WPRE) or cells expressing pCAGGS-mCherry, which were co-transfected with the untagged pcDNA-U6-shRNA, were considered to be depleted.

Postnatal electroporation. Electroporation of plasmids encoding sh1Cobl or sh2Cobl or shLrrc6 or shCtrl (U6 backbone untagged) (1.25μg) and the reporter Fig. 4 Apical actin nucleated by formins contributes to centriole stability in mature ependymal cells. a Mature (P15) ependymal cell centrioles (labelled with FOP or Cep164; red) and formins (green and grey); note that all formins colocalise with the centriolar patch (red dashed lines); cell borders are outlined with dashed white lines.b_{–d Lateral wall explants from P15 mice treated for 12 h with DMSO or SMIFH2. b F-actin (phalloidin, grey) and centrioles} (FOP, red) in mature ependymal cells from DMSO- or SMIFH2-treated explants. Cell borders and centriolar patches are outlined with dashed white and red lines, respectively. In control (DMSO-treated) explants, apical actin is enriched in the centriolar patch; formin inhibition prevents enrichment. Subapical actin staining is slightly affected by SMIFH2-treatment.c, d Ratio between mean phalloidin_{fluorescence intensities in the centriolar patch (red dashed} lines) and in the remaining cell surface (white minus red dashed lines) at the apical (c) and subapical (d) levels. Error bars represent the sem of: at the apical level, n = 96 cells for DMSO- and SMIFH2-treated explants; at the subapical level, n = 56 and 82 cells for DMSO- and SMIFH2-treated explants, respectively, from three independent experiments. P-values were determined by the Mann–Whitney test; *P = 0.0156; ***P ≤ 0.0001. Scale bars=5 μm. e Representative kymographs (200 ms) of motile cilia in P15 brain slices treated for 12 h with DMSO or SMIFH2 (formin inhibitor). The graph shows the cilia beating frequency (Hz); formin inhibition had no effect. Error bars represent the sem of n = 29, 48 cells for DMSO- and SMIFH2-treated explants from three independent experiments; P-values were determined with the Mann–Whitney test; ns, P > 0.05. f Representative images of Centrin2-GFP+centriolar patches.g Fold change in centriole number relative to DMSO-treated contralateral brain in lateral ventricles treated with SMIFH2, showing a reduction of the centriole number after SMIFH2-treatment; P-values were determined with the Mann–Whitney test. Error bars represent the sem of n = 1572 cells for DMSO- and 466 for SMIFH2-treated explants, from three independent experiments; ***P ≤ 0.0001. Scale bars=5 μm

plasmid mCherry (0.75 µg) was performed on neonates to two-day-old pups, as previously described38. This co-transfection was used throughout the paper to deplete Cobl except when co-transfected with the plasmid encoding Cobl-dsRed where sh1Cobl or shCtrl in the pTRIP backbone (1.25μg) were used.

Immunostaining. Lateral walls of the lateral brain ventricles were dissected,fixed, and immunostained, as previously described66_{. The same area of the lateral walls}

was used throughout the study (Fig.1a). The tissue was treated with 0.1% triton in

BRB80 (80 mM K-Pipes pH6.8; 1 mM MgCl2; 1 mM Na-EGTA) for 1 min and fixed in methanol at −20 °C for 10 min except for phalloidin stained tissues, which were detergent-treated and thenfixed in 4% PFA in BRB80 for 8 min. The fol-lowing antibodies were used: mouse IgG1 GT335 (1:2000, Adipogen AG-20B-0020-C100); rat anti-CD24 (1:200, BD Biosciences 557436); mouse IgG1 GTU88 (1:500, Sigma T6557); rabbit Cep164 (1:500, NOVUS 45330002); mouse IGg2b anti-AKAP450 (1:20, CTR453 gift from M. Bornens); rat anti-rootletin (1:50, Santa Cruz sc-67824); mouse IgG2b anti-Centrins 20H5 (1:2000; Millipore 04-1624); rabbit or rat anti-dsRed (1:200, Clontech 632496 or 5F8 1:400, Chromotek

5f8-Cobl-DsRed Actin Apical Subapical shCtrl sh1Cobl Centrioles Centrioles P15 Subapical actin Subapical actin Apical actin Apical actin shCtrl Centrin-GFP sh1Cobl P15 Centrin-GFP Membrane

XZ Fold change in centriole number

***

***

*

***

shCtrl 0.0 0.5 1.0 1.5 sh1Cobl shCtrl 0.0 0.5 1.0 1.5 sh1Cobl 0.5 0.7 0.9 1.1 shCtrl sh2Cobl sh1Cobl

Fold change in global

subapical actin mean intensity

(sh/surrounding cells)

Fold change in global

apical actin mean intensity

(sh/surrounding cells)

a

b

c

d

e

g

f

Fig. 5 Apical actin nucleated by Cobl also contributes to centriole stability in mature ependymal cells. a F-actin (phalloidin, grey) and Cobl (Cobl-dsRed, red) in P15 lateral ventricular walls of wild-type mice electroporated at birth with the COBL-dsRed construct; COBL-dsRed is specifically localised in the apical but not subapical actin networks. (yellow dashed lines= centriolar patch). b F-actin (phalloidin, grey) and centrioles (FOP, red) in mature ependymal cells in P15 wild-type mice electroporated at birth with shCtrl or sh1Cobl and the mCherry reporter. Cell borders of transfected cells and centriolar patches are outlined with dashed white and red lines, respectively.c, d Mean phalloidinfluorescence intensities inside the border of transfected cells relative to the mean of the three closest non-transfected control cells at the apical (c) and subapical (d) levels; apical and to a lesser extent subapical actin networks decrease in Cobl-depleted cells. Error bars represent the sem of: at the apical level, n = 35 cells for shCtrl and 46 for sh1Cobl transfected cells; at the subapical level, n = 33 cells for shCtrl and sh1Cobl from three independent experiments. P-values were determined by the Mann–Whitney test; *P = 0.0390, ***P ≤ 0.0001. e–g Mature ependymal cells on lateral ventricular walls of P15 Centrin2-GFP mice transfected at birth with shCtrl or sh1Cobl or sh2Cobl (dashed white line).e Representative images of Centrin2-GFP+centriolar patches.f Fold change in centriole number in the sh-transfected cells (shCtrl, sh1 or sh2Cobl) relative to those in surrounding cells. Error bars represent the sem of n = 236 cells for shCtrl, 129 for sh2Cobl and 106 for sh1Cobl transfected cells from three independent experiments P-values were determined by the Mann–Whitney test; ***P ≤ 0.0001. g 3D-modelling of XZ projections of centrioles (green) docked in the plasma membrane (mCherry reporter, red). Scale bars_{=5 μm in b, e and 1 µm in a, g}

100); chicken anti-GFP (1:800, Aves GFP-1020); rabbit anti-Daam1 (1:50, Euro-medex 14876-1-AP); rabbit Fhod1 (1:75, EuroEuro-medex FP3481); mouse anti-Fmn1 (1:100, Abcam AB68058); rabbit anti-mDia1 (1:200, BD Biosciences 610848); mouse IGg2b acetylated-tubulin 6-11b1 (1:600, Sigma T6793); mouse IgG2b anti-FOP (FGFR10P, 1:500, Abnova H00011116-MO) and species-specific Alexa Fluor secondary antibodies (1:800, Invitrogen). Phalloidin-A488 was used at a dilution of 1:50 with primary and secondary antibodies; Phalloidin-A568 was used at a dilution of 1:40 for the STED experiment.

Optical imagery. Images (230-nm or 210-nm z-steps) offixed samples were obtained with either an inverted epifluorescence microscope (Zeiss Axio Observer Z1) equipped with a Plan-Apochromat ×63 (NA 1.4 oil-immersion) objective, a Zeiss Apotome with an H/D grid, a CCD camera (ORCA-ER, Hammamatsu), and Zen2 software, or a confocal microscope (model SP5 or SP8; Leica) with a ×63 (Plan Neofluar NA 1.3 oil-immersion) objective and Leica LAS-AF software. Z-stack projections along the entire apical surface or 6 z-Z-stacks encompassing the apical or subapical actin networks (apical, subapical actin) are shown.

ns ns Centriole Axoneme Formins Actin Membrane

Model of centriole stabilisation by actin in ependymal cells

Immature ependymal cell Mature ependymal cell Cilia beating Detachment of centrioles Immotile / growing cilia

Cilia beating DMSO PBS + NiCl 2 1 mM P15 40 h Centrioles DMSO Cilia Centrioles D15 Supernatant 24 h Fold change in centriole number ns Fold change in centriole number shCtrl sh1Cobl P15 Flow _Flow Cilia Centrioles 0.5 0.7 0.9 1.1 0.5 0.7 0.9 1.1 0 5 10 - Apical Actin CytoD +NiCl2+flux

CytoD 2 µM

CytoD + NiCl2

Centriole+cilia number per field

***

***

CytoD 2 µm + NiCl₂ 1 mM + Flux CytoD 2 µm + NiCl₂ 1 mM CytoD 2 µm DMSO or NiCl₂ CytoD 2 µM Kif3A cKO centrin-GFP

***

CytoD 2 µm + NiCl₂ 1 mM NiCl₂ 1 mM CytoD 2 µm DMSO or PBS Kif3A cKO sh1Cobl Kif3A cKO shCtrl

a

b

d

c

e

f

g

STED nanoscopy. Images were acquired by gCW STED microscopy (TCS SP8-3 × ; Leica Microsystems) with a ×100 oil immersion objective (NA 1.4 oil-immersion); parameters were optimised for Alexa 532, Alexa 568, and Alexa 488 detection. Samples (pixel size= 30 nm) were excited sequentially at 575 nm, 530 nm, and 490 nm wavelengths with a supercontinuum laser; a 660 nm depletion laser was used. For Alexa 568, a 20% AOTF conventional scanner (400 Hz, Line Average 2, Accumulation 3) and a 60% depletion laser were used;fluorescence (585–610 nm) was collected with a hybrid detector (Gain 30%) in the gated mode (0.2–6 ns) and a 1 Airy Unit pinhole. For Alexa 532, a 20% AOTF conventional scanner (400 Hz, Line Average 2, Accumulation 3) and a 100% depletion laser were used;fluorescence (540–575 nm) was collected using a hybrid detector (Gain 30%) in the gated mode (0.2–6 ns) and a 1 Airy Unit pinhole. For GFP, a 10% AOTF conventional scanner (400 Hz, Line Average 2, Accumulation 2), and a 100% depletion laser were used;fluorescence (500–530 nm) was collected using a hybrid detector (Gain 30%) in the gated mode (0.5–6 ns) and a 1 Airy Unit pinhole. Stack images were acquired in 0.4μm Z-steps.

Phase-contrast video-microscopy. Cilia beating frequencies were analysed with an inverted epifluorescence microscope (Leica, DMIL LED) equipped with an apochromat 63 × (NA 1.32 oil-immersion) objective, a FASTCAM mini AX50 camera (Photron) and Photron fastcam software. Mature ependymal ciliary beat-ings were recorded at 250 images per second in phase contrast with transmitted light.

Transmission electron microscopy. Transmission electron microscopy was per-formed, as previously described37. Brains from P21 mice were sectioned into 200 μm slices using a vibratome (VT 1200S, Leica). Sections of the rostral lateral and dorsal walls of the lateral ventricle were taken apart in PBS under a dissecting microscope. Ventricle slices werefixed in 2.5% glutaraldehyde and 4% paraf-ormaldehyde. Tissues were treated with 1% OsO4. Fixed specimens were washed and progressively dehydrated. Samples were then incubated in 1% uranyl acetate in 70% methanol. Final dehydratations were made. Samples were pre-impregnated with ethanol/epon mix (2/1, 1/1, 1/2 ratios subsequently), and impregnated into epon. Samples were then mounted into epon blocks for 48 h at 60 °C to ensure polymerisation. Ultrathin sections of 70 nm were cut using an ultramicrotome (Ultracut E, Reichert-Jung) and analysed using a Jeol 10–11 transmission electron microscope.

Brain slices. Freshly harvested P15 brains were cut in Hank’s solution (1 × HBSS, 10 mM Hepes, 0.075% sodium bicarbonate, 1% Penicillin/Streptomycin) in 200 µm coronal sections on a vibratome (Leica VT1200S) and placed in Dulbecco’s Modified Eagle’s-Glutamax Medium (DMEM-Glutamax, Invitrogen) supple-mented with 10% FCS, 1% penicillin/streptomycin.

Drug treatments. Lateral wall explants or brain slices were immersed for the indicated times at 37 °C with 5% CO2in DMEM-Glutamax (Invitrogen)

supple-mented with 10% FCS, 1% penicillin/streptomycin and cytochalasin-D to depo-lymerise actin, SMIFH2 to inhibit formin42or NiCl2to block cilia beating39

(Sigma). The lateral ventricular wall of the opposite side of each mouse brain was used as control and immersed in the same media containing an amount of DMSO or PBS equivalent to that of the drug. Phalloidinfluorescence after a 12 h or 40 h incubation shows early and late defects in the actin cytoskeleton.

Cell cultures. Differentiating ependymal cells isolated from mouse brain were cultured, as previously described66. The shRNA constructs (0.06μg) and the Tomato reporter vector (0.04μg) were co-transfected into 0.25 × 106_{cells with the}

Jet Prime kit (polyplus transfection) when the cells were harvested for differ-entiation, according to the manufacturer’s instructions.

Supernatant collection. Ependymal cells, derived from Centrin2-GFP mice and differentiated in culture for 13–15 days, were treated for 24 h with 2 μM cytochalasin-D or DMSO with or without 1 mM NiCl2in DMEM-Glutamax

(Invitrogen) supplemented with 10% FCS, 1% penicillin/streptomycin. To generate aflow comparable to physiological values for mature cells we used the rotating unit described previously37, with the rotating plates at a height of 4 mm and a rotation speed of 6 rounds/min. Clockwise and anticlockwise rotations were used equally in our experiments.

The supernatant was then collected and Taxol (2μM) was added to stabilise cilia. The supernatant was then centrifuged for 10 min at 150 g to remove cell debris and 30 min at 12000 × g to concentrate centrioles. The pellet was resuspended in 4 ml of DMEM-Glutamax supplemented with 1% penicillin/ streptomycin and 2μM taxol and sedimented on coverslips coated with Poly-L-lysine by centrifugation at 10,400 × g for 10 min at 4 °C. The cells werefixed

in cold methanol for 10 min and the cilia stained with the GT335 antibody in PBS/ 0.1% Tween/3% BSA.

RT-PCR. Total RNA was extracted from ependymal cells with the RNeasy Midi Kit (Qiagen), according to the manufacturer’s instructions, at different time points during differentiation in vitro. RNA was converted into single-stranded cDNA using the Superscript-III Reverse Transcription System (ThermoFisher) with ran-dom primers, according to the manufacturer’s instructions. The following primers were used for subsequent PCR amplification: the housekeeping gene CyclophilinA Fw primer 5′-ACCCCACCGTGTTCTTCGAC-3′ and Rev primer 5′-CATTTGC CATGGACAAGATG-3′ for normalisation; the Cobl Fw primer 5′-TGAGATCC AAGGACAAATGG-3′ and Rev primer 5′-CTCATCTCTGATTTGGGAGG-3′. CyclophilinA and Cobl were amplified simultaneously by PCR in the same tube and the products subjected to agarose gel electrophoresis. Band intensities were quantified with ImageJ software and normalised with respect to cyclophilinA. 3D-modelling. Imaris software was used for 3D-modelling.

Data analysis. All analyses were performed on images of the rostro-dorsal part of the lateral walls of the lateral ventricles to maximise reproducibility (Fig.1a). Measurement of phalloidin_{fluorescence intensity. We compared actin} enrichment (phalloidinfluorescence) in the centriolar patch (delineated by cen-triole staining) of mature (P15) ependymal cells in ciliary mutant mice, explants treated with SMIFH2 or DMSO and Lrrc6-depleted cells. The mean intensity of phalloidinfluorescence was measured in the centriolar patch and on the surface of the rest of the cell; cell borders and actin linking the centriolar patch to the cell membrane were excluded. Apical actin was analysed on the sum projection of the 6 z-stacks encompassing Cep164 staining and subapical actin of the 6 z-stacks encompassing Akap450 staining with ImageJ software. The results are expressed as the ratio of the values in the centriolar patch over the remaining cell surface.

Total mean actin (phalloidinfluorescence) levels were measured with ImageJ software in the apical and subapical networks (as defined above) at early (P4 and P6 mice) or late (P15 mice) stages of differentiation in ependymal cells transfected with sh1Cobl or shLrrc6 or shCtrl within cell borders, in the sum projection of transfected cells normalised with respect to the mean signal intensity in the three closest non-transfected ependymal cells at the same stage, using ImageJ software. The results are expressed as the ratio of the values of the transfected cells over the mean values of the three surrounding control cells.

Fig. 6 Forces generated by cilia beating cause centrioles/cilia to detach from actin-de_{ficient ependymal cells. a, b Mature (P15) ependymal cells from} explants treated for 40 h with PBS or DMSO (Control), 1 mM NiCl2to block cilia beating, 2µM cytochalasin-D (cytoD) to depolymerise actin, or with both

drugs;aγ-tubulin stained centrioles. b Fold change in centriole number relative to DMSO- or NiCl2-treated contralateral brain, showing that cilia beating

induces centriole destabilisation in absence of actin; P-values were determined by one-way ANOVA followed by Dunn’s multiple comparison test. n = 781 cells for control-, 597 for CytoD-, 819 for NiCl2- and 547 for CytoD+ NiCl2-treated explants, from three independent experiments; ***P ≤ 0.0001; ns P

> 0.05.c, d Lateral walls of Centrin2-GFP ciliary mutant mice (Kif3A cKO) electroporated at birth with shCtrl or sh1Cobl and a reporter construct (represented as white dashed lines) andfixed at P15. c Representative Centrin2-GFP+centriolar patches.d, Fold change in centriole number relative to surrounding cells, showing no effect of Cobl-depletion on centriole number in ciliary mutant mice. P-values were determined by the Mann–Whitney test. ns, P > 0.05. n = 101 electroporated cells for Kif3A-cKO shCtrl and Kif3A-cKO sh1Cobl from three independent experiments. e Representative images of the supernatant of mature ependymal cells from Centrin2-GFP mice in culture. After 15 days of the onset of differentiation, cells were treated for 24 h with cytochalasin-D or DMSO with or without 1 mM NiCl2and an externalfluid flow; isolated centrioles (Centrin2-GFP, red) with a cilium (GT335, green) were

observed in the supernatant of CytoD-treated cultures.f Number of centrioles associated with a cilium per 63 ×field. n = 76 fields for DMSO-, 112 for CytoD-, 71 for CytoD+ NiCl2- and 61 for CytoD + NiCl2+ Flux-treated cell cultures, from three independent experiments; P-values were determined with

the Mann_{–Whitney test; ***P ≤ 0.0001. Scale bars=5 μm. g Working model: Ependymal cilia beating leads to actin enrichment at the centriolar patch.} Apical actin is in turn crucial for stabilising centrioles exposed to cilia beating andfluid flow forces. In cells with impaired apical actin, beating cilia cause centrioles and their associated cilia to detach from the cells. Error bars represent the sem in all graphs